Supported a record-breaking two-minute claim turnaround time by ensuring instant access to reliable data
Scaled self-service analytics across multiple departments, including claims, underwriting, and sales teams
Avoided additional data team headcount costs while handling increased company growth
Reduced ad hoc reporting time, freeing data scientists to build high-value predictive pricing models
Hedvig replaced its existing business intelligence platform with Looker to establish a trusted semantic layer, empowering internal teams to balance rapid sales growth with strict profitability through self-service data exploration.
Hedvig replaced its existing business intelligence platform with Looker to establish a trusted semantic layer, empowering internal teams to balance rapid sales growth with strict profitability through self-service data exploration.
Scaling insurance with a single source of truth
We needed an easier way for data to tell the same story everywhere. Looker's semantic model, and the capability for a single place to define metrics, was the key factor in choosing Looker, supporting the need for trusted metrics.
Filip Allard
Chief Pricing and Data Science Officer, Hedvig
Hedvig is not designed to play the role of a traditional insurance company. Positioning itself as the insurer for "tech-savvy people," the company caters to a demographic that manages their lives through smartphones and expects instant gratification rather than waiting months for a paper envelope in the mail. This commitment to speed is most evident in their claims processing; Filip Allard, the Chief Pricing and Data Science Officer, notes that their fastest turnaround time, from filing a claim to money hitting the bank, is just two minutes.
Yet, as is common with fast-growing companies, as Hedvig transitioned from an agile start-up to a larger organization, its internal data infrastructure began to face challenges of scale.
The company's first business intelligence platform was a basic platform, and it struggled to maintain data consistency as the number of stakeholders grew across various internal units.
The marketing team required fresh, granular data to optimize their investments and drive sales, while underwriters and pricing experts needed precise, real-time views on profitability to ensure the company was pricing risk correctly. Without a unified definition of success, these competing priorities threatened to create siloed views of the business. In the insurance world, this is a critical risk, as a sale must mean the exact same thing to a marketer as it does to an actuary or an underwriter. If one department views a policy as a success based on volume, while another views it as a failure based on the loss ratio, the company cannot scale effectively. It recognized they needed more than just a visualization tool; they needed a single source of truth.
Hedvig required a platform where business logic and metrics could be defined once in a central layer and then trusted across every department. This fundamental need for a robust, decoupled semantic model, one that could balance the tension between rapid sales growth and strict profitability, is what led them to choose Looker. By establishing this trusted layer, Hedvig ensured that every unit, from claims handling to sales, was finally speaking the same data language, allowing the organization to maintain its signature speed without sacrificing accuracy. Today, this foundation supports users who rely on these metrics to manage the delicate balance of an insurance portfolio, ensuring that growth never comes at the expense of long-term sustainability.
Dynamic data for a self-service culture
Today, Hedvig utilizes Looker alongside BigQuery, dbt, and Vertex AI to power its entire decision-making ecosystem. Their technical workflow is highly intentional: the team performs as many heavy transformations as possible, using Looker's semantic layer for the final "fine-tuning" of metrics that business users interact with. This architecture has successfully democratized data access for a growing group of users across claims, underwriting, and sales. Instead of the data team acting as a "ticket-taking" service for every minor report request, business users are now empowered to build their own dashboards and explore data independently. This shift has been transformative for the data team's efficiency, allowing them to focus on high-value tasks like building advanced pricing models.
We managed to spend less time on serving other units and more time on high-value tasks like building models. If we hadn't had Looker, then we would have needed to be a bigger team.
Filip Allard
Chief Pricing and Data Science Officer, Hedvig
A standout technical innovation in their implementation is the sophisticated use of "Liquid" functionality within LookML. Liquid is a templating language that you can use in Looker to create more dynamic content. The company describes this as a way to dynamically change aggregation levels and data sources behind the scenes in a manner that is completely seamless for the end user. For example, a user can navigate a single dashboard and switch between live data for immediate operational awareness and frozen historical data used for month-end reporting. This is achieved using native derived tables and Liquid parameters, allowing Hedvig to provide extreme detail in a way that is both fast and cost-effective. The flexibility of the tool allows Filip Allard to quickly check signals in growth and drill down into existing dashboards before needing to conduct more sophisticated analysis in Python.
This flexibility has fundamentally transformed the data team's daily life. Filip Allard notes that it is much more fulfilling for his team to build high-value predictive pricing models and refine core architecture rather than spending their days manually building basic charts for other units. If Hedvig hadn't adopted Looker, Filip Allard estimates they would have needed a significantly larger headcount just to handle the manual reporting load.
To maintain this high standard of development, the team is also exploring the use of Looker Continuous Integration and Continuous Delivery to catch errors before deployment. This ensures that their users are always working with accurate, high-performance data models that require no SQL knowledge to access.
Future-proofing for an AI-driven world
We have tweaked our data model to be more usable by agents... making them so that you require less experience to use them so that they are more self-explainable.
Filip Allard
Chief Pricing and Data Science Officer, Hedvig
With a robust BI foundation now firmly in place, Hedvig is turning its attention toward the next technological horizon: artificial intelligence and conversational analytics. Hedvig has worked extensively with ML and AI over the years in domains like pricing and claims automation. While many companies rush to implement AI on top of messy data, the Hedvig team has taken a more disciplined approach. Over the last six months, they have proactively begun refactoring their core data models to be specifically "agent-ready" for analytics. This involved them needing to revisit earlier design choices in their data structure that were overdue for fixing and cleaning up the logic to ensure that an AI agent could interpret the data without constant human intervention.
By making their data models more self-explainable and less reliant on deep "institutional knowledge," they are preparing for a future where natural language queries can empower even more users. The company is focusing on the clarity of their data model so that when an AI assistant ingests their information, the outputs are accurate and contextually aware. They want to lower the barrier to entry so far that even the least technical employee can ask a complex question about profitability and receive a reliable answer without needing years of experience with the underlying data structure. This focus on a "self-explainable" model is key to their AI strategy, ensuring accuracy as the company scales.
The team is particularly eager to explore the new updates for conversational analytics, which promise a better experience and faster latency. Additionally, the team is looking at new period-over-period LookML dimensions to further refine how they track growth trends over time. By treating their data model as a living, breathing asset that must be "taught" to be understandable by both humans and AI agents, Hedvig is not just solving today's reporting needs—they are building a future-proofed engine for an AI-driven insurance industry. This strategic focus ensures that as the world of BI evolves toward natural language and autonomous agents, Hedvig remains ahead of the curve, delivering the rapid, tech-forward experience their members expect.
Hedvig is a next-generation insurance company designed for the tech-savvy, offering instant help and lightning-fast claims handling.
Industry: Financial Services
Location: Sweden
Products: Google Cloud, Looker, BigQuery
